Custom Query (1630 matches)

Filters
 
Or
 
  
 
Columns

Show under each result:


Results (13 - 15 of 1630)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Ticket Resolution Summary Owner Reporter
#1800 fixed chicken-install fails installing eggs when encountering truncated/empty .egg-info files Vasilij Schneidermann
Description

When installing eggs with {{chicken-install}}, it checks for already installed files by walking over all installed .egg-info files with {{matching-installed-files}} and {{load-egg-info}}. If one of these .egg-info files cannot be {{read}}, the whole installation process aborts. A similar phenomenon applies to {{chicken-uninstall}} with {{remove-extension}} and {{load-egg-info}}. The user must therefore manually delete all affected egg files to have a working set of eggs again.

Two proposals:

  • Considering that the {{egg-info}} files are some sort of database, there should be a consistency check to ensure their integrity when installing/removing eggs.
  • If installing/removing eggs comes across a malformed {{egg-info}} file, it should display a warning and skip processing it. Maybe even suggest a manual uninstallation.

As for why these truncations happen, no idea. I've observed one such case with egg installation of srfi-27 and math on Windows, presumably due to maximum command length limits.

#1799 fixed [colorize egg] slow on big files curium
Description

When trying to colorize big files such as chicken's runtime.c, according to chicken profiler, 96% of time is spent on append at line 296. One enhancement could be to cons tokens and then return the reversed result. Time to colorize runtime.c goes from 10m to 02s.

Another thing which I would like is to have htmlize exported: if someone is using html-colorize, they could need that function to handle a case where the text doesn't need colorization.

Tentative diff attached.

#1797 fixed checked inline procedures should not call intern felix winkelmann sjamaan
Description

As found by Felix, inline safe procedures like C_i_char_equalp call C_i_check_char_2 which requires a call to intern0 to provide error reporting, sometimes more than once, even when there's no error.

This is highly inefficient and should be done differently. A quick fix would be to pass C_SCHEME_FALSE (or use the non-_2 version like C_i_check_char) but perhaps it's better to have C versions of the checking predicates that can call intern0 on a C string in case of error.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Note: See TracQuery for help on using queries.